You are viewing revision 1.1.0, last edited by Zack_M_Davis

In order to do inference, we constantly need to make use of categories and concepts: it is neither possible nor desirable to deal with every unique arrangement of quarks and leptons on an individual basis. Fortunately, because we don't live in a maximum-entropy universe of absolute chaos, we can talk about repeatable higher-level regularities in the world: we can distinguish particular configurations of matter as instantiations of object concepts like *chair* or *human*, and say that these objects have particular properties, like *red* or *alive*.

The sheer number of distinct configurations in which matter could be arranged is unimaginably vast, but the **superexponential conceptspace** of the number of different ways to *categorize* these possible objects is even vaster. If (for purposes of exposition) there are *n* objects in the world which either *are*, or are *not* instantiations of some concept, then the number of possible concepts is 2^*n* (for the mathematics involved, see powerset). Most of these possible concepts are complicated enough to be ruled out *a priori* by your prior; you can't expect to encounter enough evidence to cut down such a large space. The work of proper inference is to "carve reality at its joints"; to find simple generalizations and simple concepts that let you make useful inferences with respect to your goals.